Methods have been developed to assess and plot the longitudinal course of affective illness systematically in relationship to pharmacological interventions and psychosocial stressors. We have written a manual for life charting and have computerized various aspects of it. We have characterized a course of illness systematically in this fashion in more than 254 patients with primary affective disorders. These data redocument the early observations that the course of illness is often not only one of recurrences, but of a trend for progressive increases in frequency of cycling and severity of illness in refractory patients. Ultra-rapid and ultradian cycling patterns have also been identified and systematically documented for the first time. This cycle acceleration is occurring in the context of a decreasing incidence of psychosocial stressors precipitating affective episodes; i.e., as in sensitization and kindling, they are occurring autonomously. It is against this backdrop that pharmacological intervention must be considered. We have noted that one of the previous traditional interventions for bipolar depression, tricyclic antidepressants (TCAs), appear causally related to switches into mania in 35% of patients. In those individuals who show this induction, we have observed a pattern of increased rapidity of cycling in the year prior to NIMH admission and longer hospital stays at NIMH, suggesting that TCA-induced switches are a marker if not the cause of a more rapid cycling course of illness. The phenomenon of lithium discontinuation-induced refractoriness has been described in 14% of our patients. This is an additional rationale for continuing patients on long-term maintenance treatment. We have begun to assess neurobiological correlates of course of illness using a variety of neurotransmitter and endocrine markers as well as peptides and proteins in the CSF. CSF C-GRP and N-CAMs are elevated in bipolar patients, as is CSF total protein, particularly in bipolar I males. In patients studied after a prolonged period of medication-free evaluation we observed that rapid cyclers tend to show higher T4 and free T4 levels compared with non-rapid cyclers. These data contrast with previous findings in the literature in medicated patients in whom rapid cycling was often associated with hypothyroidism. Rapid cyclers respond poorly to either lithium or carbamazepine monotherapy, but 50% respond well to the combination. Differential response to pharmacological agents as a function of course of illness will be explored within the NIMH-Stanley Foundation Bipolar Research Network.